VOL. 11, NO.

9, MAY 2016 ISSN 1819-6608

ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

HIGH ALTITUDE OPERATIONS WITH PISTON ENGINES POWERPLANT

DESIGN OPTIMIZATION PART IV: RADIATORS OPTIMUM DESIGN
Luca Piancastelli, Leonardo Frizziero, Simone Pica and Giampiero Donnici
Department of Industrial Engineering, Alma Mater Studiorum University of Bologna, Viale Risorgimento, Bologna, Italy
E-Mail: leonardo.frizziero@unibo.it

ABSTRACT
In high altitude operations, the cooling system takes part to the vehicle design optimization process. An integrated
design of the cooling ducts is strictly necessary. At high altitudes, the cooling air is taken from high-pressure areas into an
alternate, extremely optimized, path. A diffuser reduces the airspeed and increases pressure of the cooling air. Then a
group of high performance finned radiators rejects the heat from coolant, air charge and oil. The high altitude, after diffuser
radiator performance is discussed in this paper. At first high performance Formula 1 radiators are introduced and discussed.
Experimental data are also exposed and summarized. The pressure drop and heat rejection are expressed in function or Re
and Pr numbers of cooling air. Then the radiator performance at high altitude is extrapolated from the ground test data.
Finally a few suggestions on radiator and cooling ducts arrangement are introduced.

Keywords: optimization, HALE, UAV, radiators, cooling, Meredith effect.

INTRODUCTION In small piston engines, intake pressurized with

Propulsion system of UAVs designed to fly
subsonically >20,000m (65,000ft) for several hours
requires very accurate design of the cooling system. In this
flight regime, Turbochargers (TC), intercoolers and
aftercoolers are needed to supply most of the intake
pressurization required to compress the small air density
into the engine. Volume flow requirements increase with
altitude, which translates to larger TC size. Pressure ratio
requirements also increase with altitude, which translates
to more TC stages. Since power is proportional to airflow
for any air breathing engine, the power plant size required
to process airflow for a rated power will grow. High
interstage and afterstage heat rejections are added to
coolant and oil ones.
Because of the increased size and weight of the
air handling system and thrust delivery components, a
propulsion system optimized for high altitudes is
significantly larger and heavier than its low altitude
counterpart. Moreover, the HALE (High Altitude Long
Endurance) vehicle needs more power to stay aloft. In fact,
faster flight speeds are necessary to maintain dynamic
pressure and support its weight in low-density air.
Therefore, the propulsion system claims greater fractions
of the airplane's gross weight. This runs counter to the
airplane's ability to carry the necessary fuel weight and
payload. The huge cooling system requirement supply
energy for an additional thrusting device: the Meredith
cooling duct. This is a subsonic ramjet engine that uses the
heat dissipated by the main propulsion system for jet
propulsion. This duct is composed by a diffuser, a radiator
system that replaces the combustor and a nozzle. The
third part of this paper introduced a simplified method for
a preliminary design of the diffuser. In this fourth part of
the paper, the radiators are introduced. A method for the
preliminary design of this heat rejection system is
discussed. At the end of the preliminary design CFD
(Computational Fluid Dynamics) is necessary to optimize
the Meredith duct.
two or three cascaded stages of turbocharging is a best
choice for stratospheric aircrafts. This paper presents the
radiator system and the duct optimization that is critical
due to low air density. Unsurprisingly, much high altitude
piston engine application failed to poor design of the
installation. Powerplants and their cooling system have
always been a problem. NACA people used extensively
their wind tunnels and their knowledge to solve cooling
problems even during the apogee of piston engines
(WWII). Many papers come from that period to revive the
knowledge of cooling that is periodically lost by the
designers. This paper describes the solutions and the
updates of this last 50 years of extensive work and
optimization of automotive cooling systems. These
updates can be directly applied to high altitude flying with
a few corrections.
The mission
Since the fuel consumption follows a cubic low
with speed, long endurance requires flying at reasonably
low speeds. The dynamic pressure available limits the
minimum speed to about 0.4M. A more likely speed will
be between 0.6 and 0.85 M to avoid excessive wingspan
and too low wing loading. In fact the aircraft should climb
through the troposphere with its climatic problems to
reach the calmer stratosphere.
Therefore, the aircraft will be more like a
sailplane than a powered aircraft and will face handling
problems at take off and lower altitudes. These problems
will be amplified by the installation of radiators and
cooling ducts.
THE COOLING CHALLENGE
The cooling drag of piston engines at high speed
and altitude may make project fail if not properly handled.
A radiator or finned barrel directly exposed in a free
airstream will not achieve efficient cooling.
In fact, less than one third of the air arriving in
front of the radiator will effectively flow across the core.

5650
 VOL. 11, NO. 9, MAY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

The rest will flow around the obstacle, not without violent
turbulence.
The efficiency is poor and the drag is prohibitive.
A duct is then strictly necessary. In order to cool, air must
flow through a radiator, with a mass flow easily
determined by calculation. It is the pressure difference
between the two faces of the radiator core, which forces
air to flow through it. Without this pressure difference, no
flow passes through the core and no cooling takes place.
The drawback is that pressure drop implies higher pressure
on the front face, and lower pressure on the rear face. The
resulting rearward force corresponds to a drag. Therefore,
cooling implies friction between the air and the radiator
core walls, and pressure difference between the faces of
the radiator.
The optimization process consists of cooling
while minimizing the parasitic drag. It is always possible.
In fact, it is difficult to cool an engine with minimal drag.
Airplane designs often combine prohibitive drag with
insufficient cooling.
High altitude cooling requirements
The Strato2C and Condor high altitude aerial
vehicles demonstrated the feasibility of the concept and
the high development costs of these solutions. Table-1
summarizes the cooling requirements of the Strato 2C
powerplant,

Table-1. Strato 2C cooling system.

Engine Engine+
Coolers LP air IP air HP air LP oil Total
coolant Gearbox oil
# of
2 2 1 1 1+1 1 9
radiators
Heat
rejection 49 53 43 90 55 10 310
(kW)
Efficiency
74 72 69 64 86 86 -
(%)
Radiator
Surface 0.65 0.76 0.6 0.8 1.1 0.12 4
(m2)

It is obvious that the enormous amount of cooling

requires a design that transforms heat energy into thrust at
least in cruise conditions. Cooling system optimization is
then of paramount importance. Power

Radiators and cooling system design 5%

Exhaust
6%
In a common approximation, the heat transfer
30%
between the radiator core and the air is proportional to the Friction
18%
air density, to the friction coefficient, to the mean air
velocity across the fins (i.e. the volume flow), and to the Coolant
temperature difference T between the air and the fins or 5%

the liquid in the radiator.

The power needed for cooling is proportional to
the square of the airspeed across the fins, and inversely
proportional to the temperature difference T between the
air and the core or cylinder fins. It is independent of the
density of air.
From this simplified theory convection cooling is
economical only at low velocity, with temperature
differences as high as possible.
A core with sufficient area (volume flow), an
airstream considerably slower than the aircraft airspeed,
and a high T minimize the internal drag.
Therefore, it is advantageous to run the engine as
hot as compatible with safe operating limitations. F1
racing engines had temperature of coolant up to 145 DEG
C (290F) and lubricant temperature of 150 DEG C.
lubricant
36%
Unburnt
Figure-1. Power cooling naturally aspirated racing car.
This result is possible with pressurized systems
that are prevalent in race applications. A pressurized
cooling system utilizes an accumulator. This accumulator
has a predetermined air room which acts as an air spring.
It also incorporates the PRV (Pressure Relieve Valve)
which is adjustable and substitutes the radiator cap.
The air spring allows the temperature expansion
to compress the air without lifting the PRV. The PRV
setting, the volume of the air spring, the coolant volume,

5651
 VOL. 11, NO. 9, MAY 2016
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
and the amount of pressure added at ambient temperature
determines the maximum coolant temperature allowed
without opening the PRV.
Once the diffuser is designed with the outmost
possible pressure increase and efficiency the radiator can
be designed. It is convenient to adopt automotive racing
heat exchanger cores.
Finned radiators
Many WWII radiators had a honeycomb matrix:
it consists of an array of copper/brass hexagonal tubes,
flared at the ends and tin soldered together (Figures 2 and
3). The air flows lengthwise through the tubes, whereas
the coolant flows around them. This core is simple to build
and offers the same wall area to the air and the liquid. This
type of core can still be found in oil radiators in marine
and industrial diesels and old jet/turboshaft engines.
However, heat rejection optimization requires that the
surface in contact with air should be considerably larger
than that offered to the liquid.
Figure-2. Honeycomb radiator cores.
Figure-3. Hawker Hurricane MKII coolant radiator, the
circular hole is for the oil radiator.
Therefore, modern radiator cores are of the finned
flat tube type. The coolant flows in the tubes with fins to
increase the air-side exchange area. End-tanks supply the
coolant to the coolant tubes.
The Reynolds number in the air passages should
be low for large heat transfer. The passage cross-section
through fin should then be small.
Surface roughness in the passages increases the
pressure drop but does not increase significantly the heat
transfer. On the contrary, properly shaped fins improve
convection.
Finned automotive radiators and air charge coolers
(aftercoolers and intercoolers) are the most cost-effective,
because of the existing technology base of mass produced
ISSN 1819-6608
automotive units that can be adapted to build a high
altitude cooling system. Recent trends in automobile
manufacturing tend to reduce weight, size and drag to
improve fuel economy. Therefore, the new radiators have
rendered the automotive technology base even more
convenient for aircraft propulsion, to the extent that many
manufacturers use automotive-derived heat-exchangers
that are extremely weight competitive even in general
aviation aero engine installations. The autoracing
marketplace for turbocharged engines already includes a
number of small business developers who mainly modify
and assemble hardware manufactured by others. The
cooling system is then cheaper because it is built up from
mass produced components.
Automotive radiator sizing
A classical method for calculating car liquid
radiator size comes from widespread knowledge of
1960s/1970s American car magazines. Several
corrections were added through time. This method applies
to spark ignition naturally aspired engines with an
efficiency around 25%. For turbocharged CRDID
(Common Rail Direct Injection Diesel) the equation also
applies. The CRDID efficiency is 38% that is typical of
Euro 3 or Euro 6 with SCR (Selective Catalytic
Reduction) vehicles. For every cc of engine displacement
add: 2 cc of core, 0.1 cc for (old) vertical flow radiator
core, 0.1 cc for an in-line engine, 0.1/0.2/0.4 cc for a
small/ medium/large trailer towing, 0.1 cc for a 2 row
radiator, 0.2 cc for outside temperature (OAT) of 105F
(40.5C), 0.2 cc for a small engine fitted to a heavy car
(modern non sportive cars), 0.2 cc for a radiator fan area
less than 78% of radiator frontal area, 0.3 cc for air
conditioning, 0.3 cc for no fan shroud, 0.3 cc for a car
with small enclosed engine compartment (modern cars),
0.5 cc for intercooler/aftercooler mounted in front of the
liquid radiator. Then it is necessary to subtract: 0.1 cc for
remote transmission cooler (not within radiator), 0.1 cc for
standard in-line transmission, 0.1 for a single row radiator,
0.2 for a spacious open air engine compartment, 0.2cc for
OAT less than 90F (32.2C), 0.4 for a large engine for
performance car. For example, the VW Polo 1043cc
radiator (1983-1990) perfectly fits. In fact, the core of the
Polo radiator is 4,000cc and the method outputs 3,963cc
(in-line engine, large trailer towing, two row radiator,
OAT of 40.5C, small engine-heavy car, small radiator fan
area, air conditioning, small enclosed engine
compartment). The method applies also for 1211cc Rotax
912 "standard radiator" that has a core of 2178 cc. In this
case, the factors of high OAT, spacious engine
compartment and the large performance car were
considered. The Rotax radiator proved to be too large for a
"high performance" application on a "sporty" ultralight
aircraft and well design cooling duct. In this case the
radiator is reduced by 36% (1323cc). The reason can be
seen in Figure-4.
5652
 VOL. 11, NO. 9, MAY 2016
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
Figure-4. Influence of external elements on radiator
performance [1].
Typically, the "design speed" of an automotive
radiator is 40 mph (62 km/h) with the engine near the
maximum torque and the fan not activated. The Take-Off
velocity of an ultralight aircraft is 15 knots more than the
stall velocity. The maximum stall velocity is 65 km/h. The
minimum TO velocity is then about 90km/h. From Figure-
5 the installed automotive radiator at 62 km/h has an SD of
310. The same radiator at 100km/h in baseline
configuration has an SD of 540. So the ultralight aircraft
can theoretically use a radiator that is 310/540=0.57 or
43% smaller in volume than a car. However, since the
aircraft acceleration during TO is not instantaneous, the
radiator volume reduction is inferior. Moreover, a slight
over-dimensioning is necessary for low-speed/max-rate
climb. Therefore, the cooling power is increased by 15%.
In this case the aircraft reduction is expressed by equation
(1).
Vaircraft Vautomotive f reduction f c lim b
f reduction f c lim b 0.57 *1.15 0.65
From these considerations the aircraft radiator is
reduced by 35%. SD is a very common way to assess the
cooling radiator performance. SD testing can be conducted
under both stable and slowly changing operating
conditions and is relatively insensitive to changes in
ambient and coolant temperatures. SD can be expressed in
terms of the heat exchanger effectiveness which is
defined as the ratio of the actual heat transfer to the
maximum possible heat transfer rate. The maximum
possible transfer rate C is always referred to the air-side
heat capacity rate (i.e. equal to the air mass flow rate
multiply by the specific heat of air) (3). The radiator
Specific Dissipation (SD) is defined by equations (2) and
(3).
mc c pc Tci Tco ma c pa Tai Tao
SD Tci Tai Tci Tai ma c pa
Where is the heat exchanger effectiveness.
ISSN 1819-6608
www.arpnjournals.com
Q
SD C ma c pa (3)
Tci Tai
Alternatively, SD is defined as the heat
dissipation rate of a radiator divided by the overall
temperature difference across the radiator (5). SD is
relatively insensitive to changes in ambient and coolant
temperatures and the tests are conducted under both stable
and slowly changing vehicle operating conditions. The
specific SD (SDs) is the SD for each unit of radiator
frontal surface. SD depends on air velocity and water
velocity. In racing radiators the coolant velocity is the
maximum possible to keep the flow laminar.
Tuning
The main source of heat in a power generation
system comes from the power the engine is producing.
This paper considers only optimally tuned engines. In fact
an un-tuned engine can place a massive load on the
cooling system even outputting less power.
Advanced/retarded timing and lean air/fuel mixtures will
require more cooling system than a properly timed engine.
Also, restrictive exhausts or improper valve timing may
have the same effect. When considering the radiator heat-
rejection capacity requirement, the engine output that you
are using continually must be examined, not the maximum
amount of power the engine can produce. This is typically
the case of sporty high-power bikes or cars engines that
will have undersized cooling systems (radiators or fins),
when used in aircraft applications. Even in many aircraft
applications, the full power-low speed cooling capacity
only needs to be enough to cool the engine while idling
and the few seconds of ground run.
(1) Radiator selection
Copper and Brass radiators were used for many
years in aircrafts and race application. Aluminium alloy is
now the preferred material. In fact, copper have a higher
thermal conductivity than aluminium. Unfortunately, the
copper fins are paired with a brass tube and lead solder.
The poor conductivity of lead solder reduces the heat
transfer rate from the tubes to the copper fins. On the
contrary aluminium alloy core is brazed. Therefore, the
tube and fin are of the same material and are brazed into
one thermally homogenous part. In addition, the
aluminium core weighs 30% of an equivalent copper/brass
core. Finally, customizing these radiators with TIG
welding is easier and more readily achieved. Radiator
thickness to frontal surface area and fin/tubes density and
geometry are subject of continuous optimization. Racing
radiators are becoming substantially thinner than in years
past.
(2)
5653
 VOL. 11, NO. 9, MAY 2016
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
Figure-5. FPI and tube spacing.
In fact, doubling a radiators available frontal
surface area doubles the available heat rejection, while
doubling the thickness of a core does not.
Figure-6. Formula 1 double pass radiator.
Fin pitch (Fins Per Inch-FPI) and tube spacing
are fundamental parameters for core heat rejection and air
side drag. FPI increase and tube spaces reduction increases
surface area and heat rejection but increases also air-side
drag. Tube spacing decreases also water-side pressure
drop through the core. Top automotive applications keep
the FPI around 15-18, while racing application currently
arrive at 25 FPI. Not professional off-road applications
tend to loosen the fin pitch down to 12 to allow dirt and
mud to pass through the core. However, it is far better to
use a proper grill to block any debris instead of less FPI.
Tube spacing and geometry has less of an impact that FPI
on the air-side efficiency of the core. In general, closer
tube spacing withstands the higher pressures that are
necessary for high altitude flight. Old single-pass radiator
had an inlet and outlet on opposite tanks and no internal
baffles. In this way the water passes through the radiator
once. There were two styles of single pass radiators, down
flow and cross flow. Crossflow (horizontal) radiators have
the end tanks on the left and on the right, while the old-
fashioned downflow (vertical) radiator had end tanks on
the top and bottom. The double pass radiator has a baffle
in between, so that the water must pass through the
radiator twice in this configuration. Therefore, the inlet
and outlet are located on the same side. Finally, there are
radiators in which the water passes through the core three
times. The inlet and outlet are on opposing corners, with
two baffles, one in each tank. The triple pass radiators are
rarely used because they have a high pressure drop on the
ISSN 1819-6608
water side. In double passing radiators the cores tube
length is effectively doubled and the height is cut in half.
Water-side backpressure is also increased. Typically going
from a single pass to a double pass radiator the heat
transfer capability increases of approximately 7%. This
achievement has virtually eliminated single pass radiators
for racing applications.
Tubes and fins can have many different shapes.
Cooling tanks shape optimization is another very common
issue in radiator design. With polymeric tanks, their shape
is highly variable without adding cost or varying the core
shape. The most common are the oval flat tubes. However,
also other shapes are available: oval, flat, ovoidal, rounded
rectangular, B-shapes, airfoil. Fins are usually multi
louvered fins that are carefully and precisely aligned to
maximize airflow and may have additional design features
to improve turbulence.
Pressure drop
A pressure drop of a modern Formula 1 racing
radiator is given by equations (4) and (5). This cross-flow,
double-pass radiator has a 1.35 (27mm) thick core with
25 FPI.
p k L Re1.6637 (4)
k L 0.1097 (5)
For radiators the Reynolds number is calculated
with the radiator fin pitch as characteristic length (6).
prad v
Re (6)
In this case prad0.0011 (25 FPI).
If the thickness of the radiator is increased
equation (4) becomes (5):
1.33
s
p k L Re 1.6637
(7)
s0
With s0=27mm. The use of the Reynolds number
instead of the velocity solves many of the design problems
due to the thinner air at 20,000m. The pressure drop
depends mainly of the FPI parameter, and, for this reason,
it is similar for topmost quality radiators.
Heat rejection
In general Q is expressed by equation (8).
Q (Tci Tai ) S (8)
But the overall external (air) heat transfer
coefficient depends on Re, Pr and on the radiator
geometry (8). It is then possible to express in function of
Re and Pr numbers for a defined radiator core.
5654
 VOL. 11, NO. 9, MAY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

1 spark ignition engines. High system pressure reduces also

2
f (Re,coregeometry) (9) head maximum temperature and fatigue due to low
Pr 3 pressure spots.

The Pr number is (10) The radiator in the Meredith cooling duct

In the case of AB=1, Ai/AB=0.35, =0.01 and
C p circular duct (D=Dh) the results of Table-1 comes from
Pr (10) Pellegrini's method (see part III of this paper).
k
Table-2. Exit data for a streamline diffuser at
Both viscosity and thermal conductivity k 20,000ISA+16.5DEG C (Hot Day -40 DEG C true temp.).
depend on temperature, while Cp is nearly constant. The
dependence from pressure is negligible. For viscosity , Vi (Mach) 0.4 0.7 0.9
the Sunderlands equation is a good approximation (11): pB[Pa] 5816 6788 8043
3 TB[K] 232 234 241
T0 C T 2
B[Pa] 0.088 0.1 0.116
0 (11)
T C T0 VB[m/s] 40 61 68
0.96 0.95 0.96
In standard air T0=291.5, 0=18.27x10-6 and
C=120 are valid values. For k a valid interpolating r 1.06 1.24 1.47
polynomial is (12)
For vi=0.4 Mach we have the following results:
k -3.68312 10 (-2682.25 T) (-1.4221 T) (12).
-8 Re=235, k=0.02, =1.51E-5, Pr=0.73. These data are not
far away from the radiator design point. In fact, the Re
number is similar to ground applications. This is due to air
It is then convenient to write equation (7) as a velocity VB=40 that is an order of magnitude larger than
function of Pr and Re numbers. The global heat transfer the design velocity of the radiator at sea level.
coefficient for the radiator of equation (4) is then (13) Therefore, the Formula 1 radiators perform well
even at high altitudes. The pressure drop is 97 Pa. This
- 1.619E - 5Re 2 0.018Re 0.877 value seems to be low but it is about 30% the pressure
2
(13) recovery of the diffuser. This is unacceptable for an
Pr 3
efficient Meredith duct. Thinner radiators should then be
used. In fact, the P51 Mustang experience demonstrated
Equation (13) holds for Re {180,580} and air that it is convenient to have a single duct with increasing
velocity between 3 and 9 m/s. The design point is Re=350, radiator temperatures. In fact, the P51D had two ducts for
Pr=0.72 and v=5.5 m/s. Air side is the most critical side of the main liquid cooler and the oil cooler, while the P51H
the radiator. Increasing the thickness of the radiator will had a single duct with the after cooler-oil (front) and the
increase the global heat transfer coefficient in a non liquid radiator (back) stacked together. In our case we
proportional way (14). have at least three different temperature levels: the after
cooler (about 120 DEG C), the coolant (140 DEG C), and
0.88
-1.619E - 5Re2 0.018Re 0.877 s the oil (150 DEG C). It is convenient to stack the three
2
(14) radiators together in order to achieve the maximum
Pr 3 s0 possible air temperature after the radiator.

Coolant flow
As velocity increases, so does the coolant
turbulation. The turbulence increases the heat rejection.
Coolant flow rate and core size are optimized to
compromise maximum heat rejection with pressure drop
and power requirement.
Static pressure throughout the system drops due
to different restrictions at each location in the system.
Therefore, system pressure is higher before the core and
lower afterwards. These pressure differentials also happen
throughout the engine due to the changes in water passage
cross section. The cylinder head is the location that needs
the highest pressures to reduce boiling risks. Maximizing
system pressure will then reduce risks of detonation in

5655
 VOL. 11, NO. 9, MAY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

vB B
vafteraircooler vcoolant (22)
coolant

Tcoolant Tcoolant Tair _ coolant

Coolant
(23)
30%

49%
The thickness of the coolant radiator can then be
Oil
evaluated (24).

21% Qcoolant Re coolant , Prcoolant , thcoolant Tcoolant S (24)

Intercooler

This process is repeated also for the oil radiator

and is iterated after the right combination of surface S and
total pressure drop is reached. The thicknesses of the air,
coolant and oil radiator can then be evaluated. Results for
Figure-7. Strato2C cooling system. Heat rejected by the cooling duct of part III and various aircraft speeds
the radiator groups. (Mach) are summarized in Table-3. In Table-3 the
maximum temperature to be cooled is 89 DEG C. This is
If the heat rejection is the one of the strato2C, the the nominal temperature for a car powerplant that does not
relative radiator thicknesses of the cooling pack will be have the oil cooler. Therefore, the values calculated in
given by equation (15) Table-3 are the minimum ones. In fact, the maximum oil
temperature for a racing car lubricant can be as high as
thk thair thcoolant thoil (15) 150 DEG C. This temperature has a very large influence of
Merediths duct thrust.
The total thickness thk is a design choice that
Table-3. Exit air data from the radiator (station 3) for
derives from maximum admissible pressure drop.
various duct inlet airspeed (Mach) (see Table-2).
Optimally and theoretically, it should not exceed 1/3 of the
diffuser pressure recovery. In this way the Meredith thrust Vi (Mach) 0.4 0.7 0.9
is optimized. In the hypothesis fo a single air charge
radiator (aftercooler only), the unknowns are the radiator T3[K] 411 372 352
surface S, the air, coolant and oil radiator thicknesses. The p3[K] 5725 6570 7721
evaluation of these unknown requires an iterative process. 3
3[kg/m ] 0.48 0.06 0.08
In fact the coolant and oil radiator are cooled by air with
different temperature, pressure and velocity than the air V3[m/s] 72 99 103
charge radiator (16)-(25). p_loss[%] 26 16 12

Tair Tair TB (16) The pressure loss p_loss factor is expressed by

equation (25).
Qair Re diffuser , Prdiffuser , thair Tair S (17)
p p0
p _ loss 1 3 100 (25)
The cooling air is heated up to Tair_coolant
p B p 0
temperature and slowed down to vafteraircooler velocity. The
Pr and Re number of the air cooling the coolant radiator
are also evaluated (18)-(22). Fuel heating
A well known problem in flight is icing. It
happens on wings, cowlings, propellers, fins and probes
Q air
T add (18) and also in filters and valves. The icing of the external
C pvB B S
_ by _ air
parts of the aircraft it typical of the climbing to the
stratosphere and to the descent phase, while the internal
fuel icing takes place during the whole flight.
Tair _ coolant TB Taddbyair (19)
The availability of heating energy should be not
overseen. Rolls Royce Trent engines use the oil cooling to
pcoolant p B pair (20) heat the fuel. In the same way it is possible to use the
cooling system of the engine to heat the fuel and the
external surfaces of the aircraft. The fuel is also a heat sink
p coolant for the aerial vehicle. In fact, during the climb phase,
coolant (21)
Tair _ coolant R where speed is low and the power output is at its

5656
 VOL. 11, NO. 9, MAY 2016
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
maximum, the fuel tanks are full. It is easy to add liquid-
to-liquid heat exchangers in strategic points of the aircraft
to dissipate energy or to increase fuel temperature. These
exchangers are small, lightweight and readily available.
They are already used for oil-coolant heat exchange. It is
also possible to use heated fuel to prevent icing on
external parts. For example, in the design phase it is
possible to design integral aluminium alloy fuel tanks on
wing leading edge.
Take off
During the take off engine thermal shock may
take place due to sudden opening of the thermostatic
valve. For this reason thermostatic valves are not common
in spark-ignition aircraft engines. However, in high
altitude aircraft overcooling may take place. A system to
close part of the radiators should then be devised.
Idling
When the engine is idling during taxiing or
waiting for the take-off clearance, the cooling system may
overheat. This is common in spark ignition engines. For
this reason fans are used in liquid cooling engines. These
fans are highly detrimental of Meredith duct efficiency
due to increased drag in cruise. For this reason it is better
to use the fuel heat sink for this purpose. Diesel engine
faces the overcooling problem. For this reason it is strictly
necessary to use a thermostatic valve in CRDIDs
(Common Rail Direct Injection Diesel).
CONCLUSIONS
In high altitude operations, the cooling system
takes a very important part to the vehicle design
optimization process. An integrated design of the cooling
ducts is strictly necessary to convert the wasted energy
into additional thrust. At high altitudes, the cooling air is
taken from high-pressure areas of the aerial vehicle into an
alternate, extremely optimized, path. A diffuser reduces
the airspeed and increases pressure of the cooling air.
Then a group of high performance finned radiators rejects
the heat from coolant, air charge and oil. The high altitude
radiator performance is discussed in this paper. The
radiators are always placed in the Meredith duct after the
diffuser. At first, high performance Formula 1 radiators
are introduced and discussed. Experimental data are also
exposed and summarized. The pressure drop and heat
rejection are expressed in function or Re and Pr numbers
of cooling air. Then the radiator performance at high
altitude is extrapolated from the ground test data. Finally a
few suggestions on radiator and cooling ducts arrangement
are introduced. A correct radiator sizing makes it possible
to obtain a positive thrust from the cooling system.
REFERENCES
[1] Wolfer H. Ignition Lag in Diesel Engines VDI-
Forchungsheft 392, 1938, (English Translation, RAE
Farnborough, Lib. No 359, UDC 621-436 - 047,
Ig5g).
ISSN 1819-6608
[2] E Mancaruso, S. S. Merola, B. M. Vaglieco. 2008.
Study of the multi-injection combustion process in a
transparent direct injection common rail diesel engine
by means of optical techniques. International Journal
of Engine Research (Impact Factor: 0.52). 01/2008;
9(6):483-498. DOI: 10.1243/14680874JER01308.
[3] H. Bettes. 2003. Flow bench applications and
techniques, Superflow.
[4] Kalitzin G. and Iaccarino G. 2002. Turbulence
Modeling in an Immersed-Boundary RANS-Method,
Center for Turbulence Research Annual Research
Briefs, Stanford University, California. pp. 415-426.
[5] Franke R. 1982. Scattered data interpolation: Tests of
some methods. Math. Comput. 38: 181-200.
[6] Ferziger J. H. and Peric M. 2002. Computational
Methods for Fluid Dynamics. Springer-Verlag, third-
edition.
[7] Lam C. K. G. and Bremhorst K. A. 1981. Modified
Form of Model for Predicting Wall Turbulence.
ASME Journal of Fluids Engineering. 103: 456-460.
[8] Wilcox D. C. 1994. Turbulence Modeling for CFD.
DCW Industries.
[9] L. Piancastelli, L. Frizziero. 2014. Turbocharging and
turbo compounding optimization in automotive
racing. Asian Research Publishing Network (ARPN).
Journal of Engineering and Applied Sciences. ISSN
1819-6608, 9(11): 2192-2199, EBSCO Publishing,
USA.
[10] L. Piancastelli, L. Frizziero, G. Donnici. 2014. The
common-rail fuel injection technique in turbocharged
di-diesel-engines for aircraft applications. Asian
Research Publishing Network (ARPN). Journal of
Engineering and Applied Sciences. ISSN 1819-6608,
9(12): 2493-2499, EBSCO Publishing, USA.
[11] L. Piancastelli, L. Frizziero. 2015. Supercharging
systems in small aircraft diesel common rail engines
derived from the automotive field. Asian Research
Publishing Network (ARPN). Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(1): 20-26,
EBSCO Publishing, USA.
[12] L. Piancastelli, L. Frizziero. 2015. Accelerated FEM
analysis for critical engine components. Published by
Walailak Journal of Science and Technology The
Walailak Journal of Science and Technology, Institute
5657
 VOL. 11, NO. 9, MAY 2016
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
of Research and Development, Walailak University,
ISSN: 1686-3933, Thasala, Nakhon Si Thammarat
80161. 12(2): 151-165, Thailand.
[13] L. Piancastelli, L. Frizziero, G. Donnici. 2015. Turbo
matching of small aircraft diesel common rail engines
derived from the automotive field. Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(1): 172-
178, EBSCO Publishing, USA.
[14] L. Piancastelli, L. Frizziero. 2015. Multistage
turbocharging systems for high altitude flight with
common rail diesel engines. Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(1): 370-
375, EBSCO Publishing, USA.
[15] L. Piancastelli, L. Frizziero. 2015. A new approach
for energy recovery and turbo compounding systems
for high altitude flight with common rail diesel
engines. Asian Research Publishing Network. ARPN
Journal of Engineering and Applied Sciences. ISSN
1819-6608, 10(2): 828-834, EBSCO Publishing,
USA.
[16] L. Piancastelli, L. Frizziero. 2015. Mapping
optimization for common rail diesel conversions from
the automotive to the flying applications. Asian
Research Publishing Network. ARPN Journal of
Engineering and Applied Sciences. ISSN 1819-6608,
10(4): 1539-1547, EBSCO Publishing, USA.
[17] L. Piancastelli, L. Frizziero. 2015. GA based
optimization of the preliminary design of an
extremely high pressure centrifugal compressor for a
small common rail diesel engine. Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(4): 1623-
1630, EBSCO Publishing, USA.
[18] L. Piancastelli, L. Frizziero, G. Donnici. 2015.
Common rail diesel-electric propulsion for small
boats and yachts. Asian Research Publishing
Network. ARPN Journal of Engineering and Applied
Sciences. ISSN 1819-6608, 10(6): 2378-2385,
EBSCO Publishing, USA.
[19] L. Frizziero, I. Rocchi. 2013. New finite element
analysis approach. Published by Pushpa Publishing
House. Far East Journal of Electronics and
Communications. ISSN: 0973-7006, 11(2): 85-100,
Allahabad, India.
ISSN 1819-6608
[20] L. Piancastelli, L. Frizziero, E. Pezzuti. 2014. Kers
applications to aerospace diesel propulsion. Asian
Research Publishing Network. ARPN Journal of
Engineering and Applied Sciences. ISSN 1819-6608,
9(5): 807-818, EBSCO Publishing, USA.
[21] L. Piancastelli, L. Frizziero. 2015. Diesel ecu
mapping optimization for aircraft and helicopter
applications. JP Journal of Heat and Mass Transfer.
ISSN 0973-5763, 11(2): 151-167, Pushpa Publishing
House, India.
[22] L. Piancastelli, L. Frizziero. 2015. Fuel consumption
reduction and power downsize via robotized multiple
speed gearbox and automatic selection system for an
automotive application. JP Journal of Heat and Mass
Transfer. ISSN 0973-5763, 12(2): 197-210, Pushpa
Publishing House, India.
[23] L. Piancastelli, M. Forghieri, L. Frizziero, L. Chinni,
M. Cremonini. 2015. Large HSDI CR diesel engines
multiple injections and multiple swirls concept. Asian
Research Publishing Network. ARPN Journal of
Engineering and Applied Sciences. ISSN 1819-6608,
10(18): 7919-7928, EBSCO Publishing, 10 Estes
Street, P.O. Box 682, Ipswich, MA 01938, USA.
[24] L. Piancastelli L., Frizziero G. Donnici. 2015.
Common rail diesel - Automotive to aerial vehicle
conversions: An update (Part I). Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(6): 2479-
2487, EBSCO Publishing, 10 Estes Street, P.O. Box
682, Ipswich, MA 01938, USA.
[25] L. Piancastelli, L. Frizziero, A. Rotondi. 2015.
Optimum installation of a common rail diesel engine
on a classical helicopter: The UH1. JP Journal of
Heat and Mass Transfer. 12(1): 45-64, 2015.
[26] L. Piancastelli, L. Frizziero, G. Donnici. 2015. A new
concept for low inertia electric turbocompounding in
racing spark ignition engines. JP Journal of Heat and
Mass Transfer. 12(1): 1-14.
[27] L. Piancastelli, L. Frizziero. 2015. Preliminary
optimization of a common rail direct injection diesel
alternative to the AE2100 class turboshafts. Asian
Research Publishing Network. ARPN Journal of
Engineering and Applied Sciences. ISSN 1819-6608,
10(11): 4738-4747, EBSCO Publishing, 10 Estes
Street, P.O. Box 682, Ipswich, MA 01938, USA.
5658
 VOL. 11, NO. 9, MAY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

[28] L. Frizziero, I. Rocchi, G. Donnici, E. Pezzuti. 2015.
Aircraft diesel engine turbocompound optimized. JP
Journal of Heat and Mass Transfer. 11(2): 133-150.
[29] L. Piancastelli, L. Frizziero, G. Donnici. 2015. The
meredith ramjet: An efficient way to recover the heat
wasted in piston engine cooling. Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(12):
5327-5333, EBSCO Publishing, 10 Estes Street, P.O.
Box 682, Ipswich, MA 01938, USA.
[30] L. Piancastelli, L. Frizziero, G. Donnici. 2015.
Common rail diesel - automotive to aerial vehicle
conversions: An update (Part II). Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(8): 3286-
3294, EBSCO Publishing, 10 Estes Street, P.O. Box
682, Ipswich, MA 01938, USA.
[31] L. Piancastelli, L. Frizziero, G. Donnici. 2015.
Common rail diesel - automotive to aerial vehicle
conversions: An update (Part III). Asian Research
Publishing Network. ARPN Journal of Engineering
and Applied Sciences. ISSN 1819-6608, 10(14):
5823-5830, EBSCO Publishing, 10 Estes Street, P.O.
Box 682, Ipswich, MA 01938, USA.
[32] Kuechemann D., Weber J. 1958. Aerodynamics of
propulsion. Mc Graw Hill, 1958
[33] L. Piancastelli, M. Pellegrini. 2007. The bonus of
aircraft piston engines, an update of the Meredith
effect. J. Heat and Technology. 25(2).
[34] 2002. Vehicle Engine Cooling Systems: Assessment
and Improvement of Wind-Tunnel Based Evaluation
Methods, A thesis submitted in accordance with the
regulations for the degree of Doctor of Philosophy by
Eton Yat-Tuen Ng, Vehicle Aerodynamics Group,
School of Aerospace, Mechanical and Manufacturing
Engineering, RMIT University, Melbourne, Australia.
[35] G. Caligiana, S. Pippa. 2005. GradH-Correction:
Guaranteed sizing gradation in multi-patch parametric
surface meshing. International Journal for Numerical
Methods in Engineering, ISSN: 1097-0207. 62(4):
495-515.
[36] A. Ceruti, G. Caligiana, F. Persiani. Comparative
evaluation of different optimization methodologies for
the design of UAVs having shape obtained by hot
wire cutting techniques, 2013, International Journal
on Interactive Design and Manufacturing. 7(2): 63-78

Symbols

Symbol Description Unit Value

Vaircraft Aircraft cruise velocity m/s -
Vautomotive Car radiator design velocity m/s -
freduction Cooling reduction factor from car to aircraft - 0.57
fclimb Cooling reduction factor from cruise velocity to climb velocity - 1.15
mc Coolant mass flow kg/s -
ma air mass flow kg/s -
cpa,Cp Air specific heat capacity at constant pressure J/kg
Tci Coolant temperature at radiator inlet C 89
Tco Coolant temperature at radiator outlet C 80
Tai Air temperature at radiator inlet C 40
Tao Air temperature at radiator outlet C -
Q Heat rejection rate kW -
kL Pressure loss factor Pa-1
v Radiator inlet air velocity m/s -
Viscosity Pa s -
s Radiator thickness m -
s0 Reference radiator thickness m 0.027
S Radiator frontal surface m2
prad Radiator fins pitch m -
Air global heat rejection factor kW K-1 m-2 -
k Air thermal conductivity W K-1 m-1 -
0 Reference viscosity stratosphere Pa s 18.27E-6
T0 Refence viscosity temperature stratosphere K 291.5

5659
 VOL. 11, NO. 9, MAY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.

www.arpnjournals.com

C Viscosity constant K 120

Vi Aircraft velocity @ 20000m M -
pB Diffuser outlet pressure (station B) Pa -
B Diffuser outlet air density (station B) kg/m3 -
TB Temperature (station B) K -
VB Velocity (station B) m/s -
A3,AB Diffuser outlet area=radiator area m2 1
Ai Diffuser inlet area m2 0.35
Diffuser efficiency - -
r Diffuser compression ratio -
thk Radiator pack thickness m -
thair Air charge radiators thicknesses m -
thcoolant Coolant radiator thickness m -
thoil Oil radiators thicknesses m -
Tair Temperature difference aftercooler C -
Qair Heat rejection aftercooler kW -
Tair_coolant Air exit temperature from aftercooler C -
Taddbyair Air temperature increase aftercooler C -
pcoolant Air pressure after the aftercooler Pa -
coolant Air density after the aftercooler kg/m3 -
vafteraircooler
Air velocity after the aftercooler m/s -
vcoolant
Tcoolant Temperature difference between cooling air and coolant C -
Qcoolant Heat rejection required by the coolant radiator kW -
HB ,DB D3 Radiator height=diffuser outlet area m -
p3 Radiator outlet pressure (station 3) Pa -
P0 Outside air pressure 20000m Pa 5475
V3 Air velocity after radiator (station 3) m/s -
3 Air density after radiator (station 3) kg/m3 -
T3 Air temperature after radiator (station 3) K -
p_loss Total ressure drop of radiator pack over diffuser pressure recovery % -

5660